Tracking Imaging Performance Bottlenecks
I have been running FOG in a number of test cases for our organization, and training myself in it preparing for what looks like will be a complete refactoring in how we handle OS deployment. I have been very happy with the options and usability that FOG has offered. But I have noticed that with the newer kernel (3.13.1), when a computer is uploading and image it appears to be a bit sluggish on the network transfer. I end up seeing the kernel fill the RAM cache with the image, then have to pause while the network transfers the information. Maybe its just that the newer kernel is that much more optimized on the drive imaging and caching side that the network just can’t keep up. But around that time I also moved the FOG server to more robust hardware (from a Dell Vostro 420 to a Dell PowerEdge 2900). I am still doing what I consider baby steps with Linux, so I was wanting some community input on what I could do to track where any of my performance bottlenecks would be. Drive Array R/W Speed? Motherboard bandwidth? RAM limitations? CPU load? Network Ux/Tx? Unoptimized software or kernels?
Obviously there are a lot of variables there, so any and all recommendations on what you all use to check out the speed of the various parts of your Linux servers would be appreciated.
A little background on my current rig:
Dell PowerEdge 2900
2x Intel Xeon E5205 @1.86GHz
4 GB RAM
Storage Controller - PERC 6/i
OS Array: RAID1 2x250GB 7.2K RPM WRITE-BACK ADAPTIVE-READ-AHEAD
Image Storage Array: RAID10 4x1TB 7.2K RPM WRITE-TRU ADAPTIVE-READ-AHEAD
2x Gigabit Ethernet Controllers
OS: Debian 7.3 x32 (fully updated).
All known proprietary kernels installed.
No other softwares installed other than FOG.
No other storage nodes in use or set up yet.
No modifications made to OS or FOG other than those critical to Debian 7.3 FOG install.